Salivary biomarkers-assisted ultrasound-based differentiation of malignant and benign thyroid nodules

• Published in: Gland surgery Vol. 11; no. 1; pp. 196 - 206
• Main Authors: Zhao, Zhifeng, Ren, Tongxin, Zhao, Yanna, Xu, Wenjuan, Xie, Rongli, Lin, Jiayun, Li, Hongjie, Zheng, Lei, Zhang, Chihao, Huo, Haizhong, Luo, Meng, Fei, Jian, Gu, Jianhua
• Format: Journal Article
• Language: English
• Published: China (Republic : 1949- ) AME Publishing Company 01.01.2022
• Subjects: Original; papillary thyroid cancer (PTC); Saliva; free-triiodothyronine (FT3); ultrasound
• ISSN: 2227-684X; 2227-8575
• DOI: 10.21037/GS-21-864

The incidence of papillary thyroid cancer (PTC) is increasing annually. ultrasonography (US) is the current primary method for evaluating thyroid nodules; however, there have been persisting challenges in diagnosing borderline malignancies. This paper aimed to establish the differential diagnostic value of salivary biomarkers for thyroid nodules geared towards improving the efficacy of US. We recruited a total of 44 PTC patients and 42 benign thyroid tumor (BTT) patients to this study. The distribution of tumor markers and thyroid hormones in saliva and serum were compared between groups; then, uni-/multi-variate logistic analyses were used to determine the risk factors of PTC. Further, we estimated the differential diagnostic value of biomarkers in thyroid nodules, especially in borderline scenarios. Finally, a multi-index diagnostic model was constructed constituting biomarkers and US. The distributions of serum thyroglobulin (TG), salivary triiodothyronine (T3), free-triiodothyronine (FT3), and free-thyroxine (FT4) were significantly different in BTT and PTC (P<0.05); salivary FT3 was identified as an independent risk factor for PTC. By analyzing the diagnostic accuracy of various Thyroid Imaging Reporting and Data System (TI-RADS) categories, category 4A was shown to have the lowest diagnostic accuracy (48.39%) with the largest proportion (31 people, 36.05%). In 4A patients, the K-nearest neighbor (KNN) algorithm attained the highest sensitivity of 87.50% and specificity of 100.00% among the machine learning-based multi-biomarkers models. Eventually, by combing the US with the KNN-based biomarkers model, the sensitivity and specificity reached 90.91% and 83.33%, respectively. Salivary biomarkers exhibit good potential in the differential diagnosis of borderline thyroid nodules and they significantly improve the prediction accuracy of the US. Additionally, we found that salivary FT3 is an independent risk factor for PTC and may be used as a key marker for PTC diagnosis.